Overall study design
EPICOR is a prospective, multinational, observational study of adults (≥18 years) surviving hospitalisation for an ACS within 24 h of symptom onset, designed to describe antithrombotic management patterns during the index hospitalisation and after discharge. The rationale and design of EPICOR have been published previously.18 Briefly, the aim was to record real-world clinical management of ACS survivors in the acute phase (prehospital and in-hospital) and during a 2-year follow-up period, as well as short-term and long-term clinical outcomes, in a wide range of hospitals and countries in four predefined regions: Northern Europe, Southern Europe, Eastern Europe and Latin America. There were three potential opportunities for enrolment in EPICOR: (1) non-transferred patients (discharged from the hospital to which they were admitted, ie, hospital 1); (2) transferred-in patients (discharged from the hospital to which they were transferred from elsewhere, ie, hospital 2) and (3) transferred-out patients (admitted to hospital 1, having been transferred out to hospital 2, and then transferred back to and discharged from hospital 1). Patients who were transferred to a non-participating hospital and not transferred back again were therefore not included in the study. Detailed information on transfer patterns in EPICOR have been published elsewhere.19
Acute phase resource use analysis
The specific aim of the present study is to evaluate, from the healthcare payer perspective, HCRU during the acute ACS phase (prehospital and in-hospital data). This was a pre-specified secondary objective of EPICOR. Data collection during this phase was prospectively carried out by the investigator at each site, using electronic case report forms. The resource analysis is given below.
Prehospital resources (from symptom onset to hospital admission): ECG, CV medication received (fibrinolytics, antiplatelet agents and anticoagulants) and the duration of ambulance transfer between hospitals for those patients who were transferred. Transport to the initial hospital was not recorded.
In-hospital resources (including all those in-patients transferred between hospitals): laboratory tests and investigations (white cell count, creatinine, blood glucose, haemoglobin, haematocrit, cardiac markers (CK-MB and troponins)); diagnostic procedures (ECG, echocardiography, treadmill, stress echocardiography, nuclear imaging, MRI, angiographic CT scan); therapeutic procedures (resuscitation, mechanical ventilation, intra-aortic balloon pumping, temporary or permanent pacemaker, implantable cardiac defibrillator, cardiac resynchronisation therapy); interventions (any cardiac catheterisation, percutaneous coronary intervention (PCI)/stents, coronary artery bypass graft (CABG) surgery or other cardiac surgery); number (%) of patients receiving CV medication (fibrinolytics, antiplatelet agents and anticoagulants); and length of hospital stay (LOS) (overall and by hospital transfer status).
HCRU was summarised by diagnosis of STEMI or NSTE-ACS, and by presence or absence of a history of CV disease (CVD), to try to understand whether this had an impact on use of resources. History of CVD included any of the following: prior IHD (myocardial infarction, PCI, CABG, chronic angina), coronary angiogram diagnostic for coronary artery disease (CAD), heart failure, atrial fibrillation, transient ischaemic attack/stroke or peripheral vascular disease. Related p values comparing the two diagnostic groups and history of CVD yes/no for each diagnostic group were produced using the χ2 or Fisher's exact test for categorical variables and the Student t test for continuous variables.
Multivariable analysis was applied to investigate the drivers of the intensity of resource use. The four possible outcome variables of interest determined by the EPICOR Executive Committee were: (1) duration of hospital stay (modelled as log(days)); (2) in-hospital coronary revascularisation (PCI or CABG; yes/no); (3) any in-hospital therapeutic procedure (resuscitation, mechanical ventilation, intra-aortic balloon pumping, temporary/permanent pacemaker, implantable cardiac defibrillator, cardiac resynchronisation therapy; yes/no) and (4) any fibrinolytic use.
A separate forward stepwise selection process considering the identified set of candidate variables was performed for each of the four outcome variables individually (linear regression model for duration of stay and logistic regression model for other outcomes). The generated models were inspected, compared and refined to incorporate agreed changes to derive a set of final individual models. The selected variables for the four models were then reconciled and a set of common models was fitted for each of the outcomes, including all of the variables that were previously selected for any of the outcomes. The individual final models were then compared with the common models to ensure that no substantial changes to the general results and conclusions were observed.